Add new marking levels for inline differences

DigiLive · DigiLive · commit 75358da3787c · 2020-11-16T11:02:41.000+01:00
* In addition to the existing line level, this commit adds marking at
  character level, word level and no inline marking.

* PhpUnit tests added for line, word and character-level marking.
diff --git a/composer.json b/composer.json
@@ -39,7 +39,8 @@
     },
     "autoload": {
         "psr-4": {
-            "jblond\\": "lib/jblond"
+            "jblond\\": "lib/jblond",
+            "Tests\\": "tests"
         }
     },
     "config": {
diff --git a/lib/jblond/Diff/Renderer/MainRenderer.php b/lib/jblond/Diff/Renderer/MainRenderer.php
@@ -4,6 +4,8 @@
 
 namespace jblond\Diff\Renderer;
 
+use jblond\Diff\SequenceMatcher;
+
 /**
  * Base renderer for rendering diffs for PHP DiffLib.
  *
@@ -135,7 +137,7 @@ protected function renderSequences(): array
 
                 if (($tag == 'replace') && ($blockSizeOld == $blockSizeNew)) {
                     // Inline differences between old and new block.
-                    $this->markInlineChange($oldText, $newText, $startOld, $endOld, $startNew);
+                    $this->markInlineChanges($oldText, $newText, $startOld, $endOld, $startNew);
                 }
 
                 $lastBlock = $this->appendChangesArray($blocks, $tag, $startOld, $startNew);
@@ -167,6 +169,118 @@ protected function renderSequences(): array
         return $changes;
     }
 
+    /**
+     * Surround inline changes with markers.
+     *
+     * @param   array  $oldText   Collection of lines of old text.
+     * @param   array  $newText   Collection of lines of new text.
+     * @param   int    $startOld  First line of the block in old to replace.
+     * @param   int    $endOld    last line of the block in old to replace.
+     * @param   int    $startNew  First line of the block in new to replace.
+     */
+    private function markInlineChanges(
+        array &$oldText,
+        array &$newText,
+        int $startOld,
+        int $endOld,
+        int $startNew
+    ): void {
+        if ($this->options['inlineMarking'] < self::CHANGE_LEVEL_LINE) {
+            $this->markInnerChange($oldText, $newText, $startOld, $endOld, $startNew);
+
+            return;
+        }
+
+        if ($this->options['inlineMarking'] == self::CHANGE_LEVEL_LINE) {
+            $this->markOuterChange($oldText, $newText, $startOld, $endOld, $startNew);
+        }
+    }
+
+    /**
+     * Add markers around inline changes between old and new text.
+     *
+     * Each line of the old and new text is evaluated.
+     * When a line of old differs from the same line of new, a marker is inserted into both lines, just before the first
+     * different character/word. A second marker is added just before the following character/word which matches again.
+     *
+     * Setting parameter changeType to self::CHANGE_LEVEL_CHAR will mark differences at character level.
+     * Other values will mark differences at word level.
+     *
+     * E.g. Character level.
+     * <pre>
+     *         1234567890
+     * Old => "aa bbc cdd" Start marker inserted at position 4
+     * New => "aa 12c cdd" End marker inserted at position 6
+     * </pre>
+     * E.g. Word level.
+     * <pre>
+     *         1234567890
+     * Old => "aa bbc cdd" Start marker inserted at position 4
+     * New => "aa 12c cdd" End marker inserted at position 7
+     * </pre>
+     *
+     * @param   array  $oldText   Collection of lines of old text.
+     * @param   array  $newText   Collection of lines of new text.
+     * @param   int    $startOld  First line of the block in old to replace.
+     * @param   int    $endOld    last line of the block in old to replace.
+     * @param   int    $startNew  First line of the block in new to replace.
+     */
+    private function markInnerChange(array &$oldText, array &$newText, int $startOld, int $endOld, int $startNew): void
+    {
+        for ($iterator = 0; $iterator < ($endOld - $startOld); ++$iterator) {
+            // ChangeType 0: Character Level.
+            // ChangeType 1: Word Level.
+            $regex = $this->options['inlineMarking'] ? '/\w+|[^\w\s]|\s/u' : '/.?/u';
+
+            // Deconstruct the lines into arrays, including new empty element to the end in case a marker needs to be
+            // placed as last.
+            $oldLine   = $this->sequenceToArray($regex, $oldText[$startOld + $iterator]);
+            $newLine   = $this->sequenceToArray($regex, $newText[$startNew + $iterator]);
+            $oldLine[] = '';
+            $newLine[] = '';
+
+            $sequenceMatcher = new SequenceMatcher($oldLine, $newLine);
+            $opCodes         = $sequenceMatcher->getGroupedOpCodes();
+
+            foreach ($opCodes as $group) {
+                foreach ($group as [$tag, $changeStartOld, $changeEndOld, $changeStartNew, $changeEndNew]) {
+                    if ($tag == 'equal') {
+                        continue;
+                    }
+                    if ($tag == 'replace' || $tag == 'delete') {
+                        $oldLine[$changeStartOld] = "\0" . $oldLine[$changeStartOld];
+                        $oldLine[$changeEndOld]   = "\1" . $oldLine[$changeEndOld];
+                    }
+                    if ($tag == 'replace' || $tag == 'insert') {
+                        $newLine[$changeStartNew] = "\0" . $newLine[$changeStartNew];
+                        $newLine[$changeEndNew]   = "\1" . $newLine[$changeEndNew];
+                    }
+                }
+            }
+
+            // Reconstruct the lines and overwrite originals.
+            $oldText[$startOld + $iterator] = implode('', $oldLine);
+            $newText[$startNew + $iterator] = implode('', $newLine);
+        }
+    }
+
+    /**
+     * Split a sequence of characters into an array.
+     *
+     * Each element of the returned array contains a full pattern match of the regex pattern.
+     *
+     * @param   string  $pattern   Regex pattern to split by.
+     * @param   string  $sequence  The sequence to split.
+     *
+     * @return array  The split sequence.
+     */
+    public function sequenceToArray(string $pattern, string $sequence): array
+    {
+        preg_match_all($pattern, $sequence, $matches);
+
+        return $matches[0];
+    }
+
     /**
      * Add markers around inline changes between old and new text.
      *
@@ -187,15 +301,15 @@ protected function renderSequences(): array
      * @param   int    $endOld    last line of the block in old to replace.
      * @param   int    $startNew  First line of the block in new to replace.
      */
-    private function markInlineChange(array &$oldText, array &$newText, int $startOld, int $endOld, int $startNew)
+    private function markOuterChange(array &$oldText, array &$newText, int $startOld, int $endOld, int $startNew): void
     {
         for ($iterator = 0; $iterator < ($endOld - $startOld); ++$iterator) {
             // Check each line in the block for differences.
             $oldString = $oldText[$startOld + $iterator];
             $newString = $newText[$startNew + $iterator];
 
             // Determine the start and end position of the line difference.
-            [$start, $end] = $this->getInlineChange($oldString, $newString);
+            [$start, $end] = $this->getOuterChange($oldString, $newString);
             if ($start != 0 || $end != 0) {
                 // Changes between the lines exist.
                 // Add markers around the changed character sequence in the old string.
@@ -233,7 +347,7 @@ private function markInlineChange(array &$oldText, array &$newText, int $startOl
      *
      * @return array Array containing the starting position (0 by default) and the ending position (-1 by default)
      */
-    private function getInlineChange(string $oldString, string $newString): array
+    private function getOuterChange(string $oldString, string $newString): array
     {
         $start = 0;
         $limit = min(mb_strlen($oldString), mb_strlen($newString));
diff --git a/lib/jblond/Diff/Renderer/MainRendererAbstract.php b/lib/jblond/Diff/Renderer/MainRendererAbstract.php
@@ -11,17 +11,32 @@
  *
  * PHP version 7.2 or greater
  *
- * @package     jblond\Diff\Renderer
- * @author      Mario Brandt <leet31337@web.de>
- * @author      Ferry Cools <info@DigiLive.nl>
+ * @package         jblond\Diff\Renderer
+ * @author          Mario Brandt <leet31337@web.de>
+ * @author          Ferry Cools <info@DigiLive.nl>
  * @copyright   (c) 2009 Chris Boulton
- * @license     New BSD License http://www.opensource.org/licenses/bsd-license.php
- * @version     2.2.1
- * @link        https://github.com/JBlond/php-diff
+ * @license         New BSD License http://www.opensource.org/licenses/bsd-license.php
+ * @version         2.2.1
+ * @link            https://github.com/JBlond/php-diff
  */
 abstract class MainRendererAbstract
 {
-
+    /**
+     * Mark inline character differences.
+     */
+    public const CHANGE_LEVEL_CHAR = 0;
+    /**
+     * Mark inline word differences.
+     */
+    public const CHANGE_LEVEL_WORD = 1;
+    /**
+     * Mark line differences.
+     */
+    public const CHANGE_LEVEL_LINE = 2;
+    /**
+     * Mark no inline differences.
+     */
+    public const CHANGE_LEVEL_NONE = 4;
     /**
      * @var Diff $diff Instance of the diff class that this renderer is generating the rendered diff for.
      */
@@ -30,6 +45,11 @@ abstract class MainRendererAbstract
     /**
      * @var array   Associative array containing the default options available for this renderer and their default
      *              value.
+     *              - inlineMarking     The level of how differences are marked.
+     *                                  - self::CHANGE_LEVEL_NONE   Don't Inline-Mark.
+     *                                  - self::CHANGE_LEVEL_CHAR   Inline-Mark each different character.
+     *                                  - self::CHANGE_LEVEL_WORD   Inline-Mark each different word.
+     *                                  - self::CHANGE_LEVEL_LINE   Inline-Mark from first to last line diff.
      *              - tabSize           The amount of spaces to replace a tab character with.
      *              - format            The format of the input texts.
      *              - cliColor          Colorized output for cli.
@@ -40,6 +60,7 @@ abstract class MainRendererAbstract
      *              - deleteColors      Fore- and background color for removed text. Only when cliColor = true.
      */
     protected $mainOptions = [
+        'inlineMarking'   => self::CHANGE_LEVEL_LINE,
         'tabSize'         => 4,
         'format'          => 'plain',
         'cliColor'        => false,
@@ -60,7 +81,7 @@ abstract class MainRendererAbstract
      * The constructor. Instantiates the rendering engine and if options are passed,
      * sets the options for the renderer.
      *
-     * @param array $options Optionally, an array of the options for the renderer.
+     * @param   array  $options  Optionally, an array of the options for the renderer.
      */
     public function __construct(array $options = [])
     {
@@ -72,9 +93,11 @@ public function __construct(array $options = [])
      *
      * Options are merged with the default to ensure that there aren't any missing options.
      * When custom options are added to the default ones, they can be overwritten, but they can't be removed.
+     *
+     * @param   array  $options  Array of options to set.
+     *
      * @see MainRendererAbstract::$mainOptions
      *
-     * @param array $options Array of options to set.
      */
     public function setOptions(array $options)
     {
diff --git a/tests/Diff/Renderer/MainRendererTest.php b/tests/Diff/Renderer/MainRendererTest.php
@@ -1,5 +1,7 @@
 <?php
 
+/** @noinspection PhpMethodNamingConventionInspection */
+
 declare(strict_types=1);
 
 namespace Tests\Diff\Renderer;
@@ -8,23 +10,25 @@
 use jblond\Diff\Renderer\MainRenderer;
 use PHPUnit\Framework\TestCase;
 use ReflectionClass;
+use ReflectionException;
 
 /**
  * PHPUnit Test for the main renderer of PHP DiffLib.
  *
  * PHP version 7.2 or greater
  *
- * @package     Tests\Diff\Renderer
- * @author      Mario Brandt <leet31337@web.de>
- * @author      Ferry Cools <info@DigiLive.nl>
+ * @package         Tests\Diff\Renderer
+ * @author          Mario Brandt <leet31337@web.de>
+ * @author          Ferry Cools <info@DigiLive.nl>
  * @copyright   (c) 2009 Mario Brandt
- * @license     New BSD License http://www.opensource.org/licenses/bsd-license.php
- * @version     2.2.1
- * @link        https://github.com/JBlond/php-diff
+ * @license         New BSD License http://www.opensource.org/licenses/bsd-license.php
+ * @version         2.2.1
+ * @link            https://github.com/JBlond/php-diff
  */
 
 /**
  * Class MainRendererTest
+ *
  * @package Tests\Diff\Renderer\Html
  */
 class MainRendererTest extends TestCase
@@ -40,9 +44,9 @@ class MainRendererTest extends TestCase
     /**
      * MainRendererTest constructor.
      *
-     * @param null   $name
-     * @param array  $data
-     * @param string $dataName
+     * @param   null    $name
+     * @param   array   $data
+     * @param   string  $dataName
      */
     public function __construct($name = null, array $data = [], $dataName = '')
     {
@@ -86,14 +90,14 @@ public function testRenderSimpleDelete()
     /**
      * Call protected/private method of a class.
      *
-     * @param object &$object     Instantiated object that we will run method on.
-     * @param string  $methodName Method name to call
-     * @param array   $parameters Array of parameters to pass into method.
+     * @param   object  $object      Instantiated object that we will run method on.
+     * @param   string  $methodName  Method name to call
+     * @param   array   $parameters  Array of parameters to pass into method.
      *
      * @return mixed Method return.
-     * @throws \ReflectionException If the class doesn't exist.
+     * @throws ReflectionException If the class doesn't exist.
      */
-    public function invokeMethod(&$object, $methodName, array $parameters = [])
+    public function invokeMethod(object $object, string $methodName, array $parameters = [])
     {
         $reflection = new ReflectionClass(get_class($object));
         $method     = $reflection->getMethod($methodName);
@@ -137,4 +141,52 @@ public function testRenderFixesSpaces()
             $result
         );
     }
+
+    /**
+     * Test inline marking for changes at line level.
+     *
+     * Everything from the first difference to the last difference should be enclosed by the markers.
+     *
+     * @throws ReflectionException When invoking the method fails.
+     */
+    public function testMarkOuterChange()
+    {
+        $renderer = new MainRenderer();
+        $text1    = ['one two three four'];
+        $text2    = ['one tWo thrEe four'];
+        $this->invokeMethod($renderer, 'markOuterChange', [&$text1, &$text2, 0, 1, 0]);
+        $this->assertSame(["one t\0wo thre\1e four"], $text1);
+        $this->assertSame(["one t\0Wo thrE\1e four"], $text2);
+    }
+
+    /**
+     * Test inline marking for changes at character and word level.
+     *
+     * At character level, everything from a different character to any subsequent different character should be
+     * enclosed by the markers.
+     *
+     * At word level, every word that is different should be enclosed by the markers.
+     *
+     * @throws ReflectionException When invoking the method fails.
+     */
+    public function testMarkInnerChange()
+    {
+        $renderer = new MainRenderer();
+
+        // Character level.
+        $renderer->setOptions(['inlineMarking' => $renderer::CHANGE_LEVEL_CHAR]);
+        $text1 = ['one two three four'];
+        $text2 = ['one tWo thrEe fouR'];
+        $this->invokeMethod($renderer, 'markInnerChange', [&$text1, &$text2, 0, 1, 0]);
+        $this->assertSame(["one t\0w\1o thr\0e\1e fou\0r\1"], $text1);
+        $this->assertSame(["one t\0W\1o thr\0E\1e fou\0R\1"], $text2);
+
+        // Word Level.
+        $renderer->setOptions(['inlineMarking' => $renderer::CHANGE_LEVEL_WORD]);
+        $text1 = ['one two three four'];
+        $text2 = ['one tWo thrEe fouR'];
+        $this->invokeMethod($renderer, 'markInnerChange', [&$text1, &$text2, 0, 1, 0]);
+        $this->assertSame(["one \0two\1 \0three\1 \0four\1"], $text1);
+        $this->assertSame(["one \0tWo\1 \0thrEe\1 \0fouR\1"], $text2);
+    }
 }
